Field of the Invention
Aspects of the present invention relate to a technology for reducing the noise of an image signal generated by as imaging apparatus.
Description of the Related Art
Imaging devices where photoelectric conversion devices are arranged, such as image sensors, are widely used. However, a noise component may be included to reduce the quality of an image. Therefore, various noise reduction methods have been proposed.
For example, Japanese Patent No. 5337049 is knows as a method for varying a parameter of a noise reducing algorithm based on noise level in an image. Japanese Patent No. 5337049 discloses a method in which noise information is obtained as a function of coordinates in a digital image, and one or more parameters for a noise reduction process are varied based on the obtained noise information during the noise reduction process.
In recent years, imaging devices with new structures have been proposed. Even if the above noise reduction process is used, a sufficient noise reduction effect may not be able to be obtained. For example, an imaging device has been proposed which includes a plurality of pixels having a pupil-division function of an imaging optical system that forms an optical image of a subject, and enables focus detection by phase difference detection with an image signal. In an imaging apparatus using such an imaging device, signals for focus detection are not necessarily required to be read out from an entire image. The signals for focus detection may be read out only from a partial area.
However, if an area where the signal for focus detection is read out and an area where a signal for image generation is read out coexist in the same image, the amounts of noise superimposed on image signals corresponding respectively to both areas are different. The reason of it is briefly described taking one example. The example a case where one pixel includes two photodiodes that achieve the pupil-division function in the area where the signal for focus detection is read out, a signal obtained from one photodiode is output first, and then signals obtained from two photodiodes are added to be output. By contrast, in the area where the signal for image generation is read out, only an operation of adding signals obtained from the two photodiodes and outputting the resultant signal is performed. In the area where the signal for focus detection is read out, the signal output operation is performed twice; therefore, it takes a longer time between the performance of a reset operation and the completion of readout than in the area where the signal for image generation is read out and where the signal output operation is performed only once. Consequently, the operating frequency between signals is reduced and flicker noise increases. Hence, focus is detected only in a part of an image and the signal for focus detection is read out only in a partial area, there arises a problem that a noise texture changes between the area where signal for focus detection has been read out and the area where signal for focus detection has not been read out.
In the signal processing method described in Japanese Patent No. 5337049, the noise level information as a function of coordinates is used to attempt noise reduction. However, no consideration is given to a case where a different signal reading operation is applied according to different areas in the same image.